Stage 2: Architecture Design#

Objective#

Design the CLI command structure, state model, and output format based on the SOP analysis.

Inputs#

• <SOFTWARE>.md from Stage 1

Steps#

1. Design Command Groups — map each workflow to a Click/Commander command group:

   cli-anything-<SOFTWARE>
   ├── project    (create, open, save, close, info)
   ├── <domain1>  (domain-specific commands)
   ├── <domain2>  (domain-specific commands)
   ├── export     (render, compile, convert)
   └── session    (undo, redo, history, snapshot)

1. Design State Model:
   • What is the "project" concept? (file, workspace, session)
   • What state changes between commands? (selections, settings, layers)
   • Deep-copy undo/redo: each command snapshots state before execution

1. Design Output Format:
   • Human-readable: clear terminal output with color hints
   • JSON (--json): structured output for agent consumption

   {
     "status": "success|error",
     "command": "<command.subcommand>",
     "data": { ... },
     "metadata": { "duration_ms": 123, "software_version": "x.y.z" }
   }

1. Design Backend Calls:
   • Map each command to specific software invocations
   • Define subprocess timeout per command category
   • Define error detection patterns (exit codes, stderr patterns)

1. Generate architecture.json:

   {
     "commandGroups": [...],
     "stateModel": {...},
     "outputFormat": {...},
     "backendCalls": [...]
   }

Output#

• architecture.json — the complete design specification